Editorial Note
This dossier does not develop an original thesis. It gathers existing works that all pose, in incompatible vocabularies, the same question: under what condition does a posterior event modify the status of an anterior event? References are commented for what they do operationally, not for what they conclude.
1. John Archibald Wheeler: the delayed-choice experiment
Wheeler proposes in 1978 a thought variant of the double-slit experiment in which the measurement apparatus is chosen after the particle has already traversed the system. If one chooses to measure which path the particle took, it seems to have taken a defined path. If one chooses to measure interference, it seems to have taken both. The choice is made after. The particle has already passed.
Wheeler is precise about what this does not mean. No signal travels back in time. No late instruction comes to modify an already inscribed event. What changes is the admissible description of what happened. The posterior measurement does not rewrite the past event. It completes the regime in which this event becomes describable.
Kim et al. realize in 2000 an experimental version of the delayed-choice quantum eraser. What was a thought experiment becomes a measurable result. Here again, the decisive point is not that a message would be sent toward the past, but that certain patterns appear only once events are grouped according to an ulterior condition.
What Wheeler does: he distinguishes naive retro-causality and structural retro-causality. The first imagines a future that produces the past. The second describes a system whose complete description requires a terminal condition. This distinction is the starting point of everything that follows in this dossier.
References
- Wheeler, J. A. (1978). The "past" and the "delayed-choice" double-slit experiment. In A. R. Marlow (Ed.), Mathematical Foundations of Quantum Theory (pp. 9-48). Academic Press.
- Kim, Y.-H., Yu, R., Kulik, S. P., Shih, Y., & Scully, M. O. (2000). Delayed "choice" quantum eraser. Physical Review Letters, 84(1), 1-5.
2. Jorge Luis Borges: Kafka and his precursors
Borges observes that reading Kafka renders visible a family of anterior texts that had not been perceived as linked. Zeno, Han Yu, Kierkegaard, Browning, Bloy share something that no one had named before Kafka. Borges draws a conclusion: if Kafka had not written, we would not perceive this series.
The future work creates its precursors. It does not invent them; the texts existed. It creates the condition under which they become readable as a series. The series did not exist before the term that reveals it, not because the texts were missing, but because no framework yet united them.
Borges does not speak of physical causality. He speaks of retrospective readability. But the operation is structurally comparable to Wheeler's: a posterior event completes the description of anterior events without modifying them. The texts of Zeno and Kierkegaard have not changed. What has changed is the regime in which they can be read together.
T. S. Eliot formulates in 1919 a neighboring idea in "Tradition and the Individual Talent": a new work retrospectively modifies the order of existing works. The ancient order was complete before the arrival of the new work; to remain an order after this novelty, it must be slightly altered.
What Borges and Eliot do: they show that the order of a tradition is revised from its provisional end, not only from its origin. The series receives its meaning from the term that temporarily closes it.
References
- Borges, J. L. (1951). Kafka y sus precursores. In Otras inquisiciones. Sur. Tr. fr. : Kafka et ses précurseurs. In Enquêtes. Gallimard, 1957.
- Eliot, T. S. (1919). Tradition and the individual talent. The Egoist, 6(4), 54-55.
3. Charles Darwin: natural selection as retrospective teleology
Natural selection operates without intention. There is no end posited in advance, no project, no internal direction toward a determined result. Yet, once selection has operated, certain anterior variations can be reread as adaptive.
The trait did not carry its status alone. A variation is not adaptive in itself, isolated from all environment, all competition, all reproduction. It becomes so in a relation between organism, milieu and descendance. What survives permits requalifying certain anterior variations as having been favorable.
Teleology is thus an effect of reading from the result. The eye seems designed to see because one reads it from accomplished vision. The wing seems oriented toward flight because one reads it from the lineages that survived. If one read from the origin, one would see not a project, but variations, losses, constraints, bifurcations.
Daniel Dennett develops this point in Darwin's Dangerous Idea. He describes Darwinism as an algorithmic process: blind, without intentionality, but capable of producing forms that resemble solutions after the fact. Richard Dawkins, in The Blind Watchmaker, insists on the same teleological illusion: biological complexity appears designed because it is read from the result.
What Darwin, Dennett and Dawkins do: they show that a process can produce orientation without intention. The terminal condition is not here a future cause. It is the result from which the past is classified. Adaptation is not an intrinsic property of the trait. It is a relational property, stabilized after the fact by differential survival.
References
- Darwin, C. (1859). On the Origin of Species by Means of Natural Selection. John Murray.
- Dawkins, R. (1986). The Blind Watchmaker. Norton.
- Dennett, D. C. (1995). Darwin's Dangerous Idea: Evolution and the Meanings of Life. Simon & Schuster.
4. Inheritance law: the will as terminal condition
A will is a document written in the present to produce effects in a future from which the signatory will be absent. It converts a present intention into future obligation, without the testator being able to verify execution or correct it after the fact.
But the will does more than defer a will. It installs a terminal condition from which anterior acts can be counted, reported, limited or opposed at the moment of succession. It does not materially modify these acts. It changes the regime in which they will be read.
A past donation, an accorded advantage, an anterior disposition or family transmission can receive another status when the succession opens. What was an isolated act enters into a terminal ensemble: succession mass, reserve, disposable portion, report, reduction. Death does not create these acts. It closes the system in which they become calculable.
The hereditary reserve, in civil law systems, operates according to a similar logic. It fixes in advance what the testator cannot freely dispose of. It bounds the space of admissible wills before these are even formulated. The will is thus written from its limit.
Roman law already distinguished the testamentum from the simple donatio mortis causa. In both cases, death is not only a biological event. It is the juridical condition that decides the regime of the act. Before death, the will is a revocable document. After, it becomes operative.
What the will does: it materializes the terminal condition in a juridical object. It shows that this structure is not a theoretical abstraction, but an ancient, formalized, opposable institutional operation.
References
- Cornu, G. (2007). Droit civil : La famille. Montchrestien.
- Terré, F., & Lequette, Y. (2015). Droit civil : Les successions, les libéralités. Dalloz.
5. Backtesting in finance: reading the past from a known future
Backtesting is the procedure by which an investment strategy is evaluated by applying it to historical data. One takes a decision rule, projects it onto past prices, then measures the performance it would have produced. If the result is satisfactory, the strategy receives credibility for the future.
The structural problem is well known: backtesting does not only test a strategy on the past. It tests a strategy constructed from a present that already knows the sequel. The danger is not that data know the future. It is that the researcher knows it. He chooses rules, windows, exclusions, parameters, thresholds and asset universes in an already closed past.
The strategy that "worked" between 2000 and 2010 perhaps worked in a past rendered readable from 2024. The 2008 crisis, bankruptcies, survivors, missing data, disappeared assets, overly well-adjusted parameters: all this can silently enter the test construction. The past used by the backtest is not the open past of actors of the time. It is a past reconstructed from a known future.
David Hume had posed the general problem: inference from past to future presupposes that the future will resemble the past. Backtesting adds a twist: it presupposes that the tested past has not already been organized by the future of the one testing it.
Nassim Nicholas Taleb names "narrative fallacy" the tendency to reconstruct the past as if it had been more readable than it was. Backtesting can become the instrumented version of this reinterpretation: it transforms the past into proof of the future while forgetting that this past has already been reread from a future.
What backtesting does: it shows that structural retro-causality is not only a question of physics or literature. It is a daily operation in financial systems: a known future selects the forms of the past that will become arguments for a present decision.
References
- Taleb, N. N. (2007). The Black Swan: The Impact of the Highly Improbable. Random House.
- Lopez de Prado, M. (2018). Advances in Financial Machine Learning. Wiley.
- Hume, D. (1748). An Enquiry Concerning Human Understanding. Millar.
Synthesis
These five cases do not say the same thing. Wheeler works on quantum mechanics, Borges on literature, Darwin on biology, inheritance law on civil obligations, backtesting on financial markets. Their vocabularies are incompatible. Their methods are unrelated.
They nonetheless pose the same structural question.
A posterior event can complete the description of an anterior event without modifying it. The particle has not changed trajectory. Zeno's texts have not changed. The biological trait has not changed. The anterior juridical act has not changed. Historical prices have not changed.
What changes, in each case, is the regime in which these events become describable. The terminal condition is not a cause that travels back in time. It is the second boundary of a description that was incomplete without it.
The past does not change content.
It receives its second edge.
